1 / 42

Disruptive innovation, Big Data and Algorithms

Disruptive innovation, Big Data and Algorithms. Antonio Gomes 31 August 2017. Outline. Disruptive Innovation Competition and Regulation Challenges to C ompetition Agencies Big Data Algorithms and collusion. Competition , Innovation and Digital Economy.

samira
Download Presentation

Disruptive innovation, Big Data and Algorithms

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Disruptive innovation, Big Data and Algorithms Antonio Gomes 31 August 2017

  2. Outline DisruptiveInnovation Competition and Regulation Challenges to Competition Agencies Big Data Algorithmsandcollusion

  3. Competition, Innovation and Digital Economy • In 2016, the digital economy and innovation was selected as a long-term theme for discussion at the OECD Competition Committee • Four main sub-streams: • The relationship between the digital economy, competition law and innovation • Challenges posed to prevailing antitrust tools and approaches • Practical challenges to competition enforcement • Detailed industries and sectors

  4. Disruptive innovation Innovations that drastically alter markets Breakthroughs rather than incremental technological developments Radical changes that restructure or create entire markets Typically from outside a market’s value network, but incumbents sometimes disrupt their own market (e.g. Nestlé with Nespresso) New products, processes or business models

  5. Disruptive innovation • Many examples of disruptive innovation in history • e.g. assembly line manufacturing, internal combustion engine, CDs, etc • Main difference today is the pace of disruptive innovation (speed and force) • Some disrupted markets exhibit network effectsresulting in a veryrapidgrowthphase for the disruptor • Often disruptors • Bypass intermediaries • Reduceunecessarycosts • Avoidregulations/costsimposed by regulation • Typically reduces or destroys market shares of incumbent firms (e.g. Nokia displaced by Apple and Google) - resistance by incumbents

  6. Disruptive innovation • Disruptive Innovation often appears in sectors subject to (heavy) regulation • Taxis • Hotels • Financial services • Electricitygeneration (microgeneration) • Legal and other expert professionalservices • Incumbents call for applying existing regulation to new entrants even when it is not well-suited - claims of “fair competition” (e.g. Taxi driver reactions to Uber and alike services)

  7. Disruptive innovation • When disruptive innovation takes place in a highly regulated market regulation may distort competition • Disruptive innovation may render the regulation obsolete • Regulation may block, deter or retard entry by disruptive firms • Regulation can raise unjustified barriers to entry (e.g., numerus clausus restrictions) • But can also unevenly impose unnecessary burdens on established firms (e.g., licensing obligations), compromising their strategic reaction towards innovators

  8. Competition and Regulation Important to understandwhat are thepolicy objectives ofregulation Regulatorsoften look at safety/consumer protection as an absolute value – but innovationmaybealreadyaddressingunderlyingpolicyconcerns Role ofreputational feedback mechanisms (e.g. online customer ratings) vs more traditionalwaysofdealingwithriskssuch as licensingorotherdirectcontrols Regulators can use data to reducethecostsofregulation Competition andprivacyconcerns

  9. Challenges to Competition Agencies Competition Advocacy • Raiseawarenessregardingthebenefitsofcompetition • Influencelegislators, regulators and the general public • Comment onproposedorexistingregulations, opinions and recommendations • Appearing beforelawmakers to discusscompetitionmatters • Conductingmarketstudies • Disseminate information

  10. Challenges to Competition Agencies Enforcement ofcompetitionlaw • MergersandAcquisitions • Incumbentfirmsacquiring a disruptiveentrant • Ifdisruptorstillnascent, maybedifficult to establishharm • Firms not yet earning revenue may escape merger notification (use valueoftransaction as analternative?) • Horizontal agreements/Collusivebehaviour • The use of Big data and algorithms to facilitate collusion

  11. Challenges to Competition Agencies • Unilateral Conduct • Byincumbentswhichmayadoptexclusionaryconductagainst a disruptivefirm, butdisruptivefirmsoftenmanage to erode demandofincumbentsevenwhenthesebecome more price-competitive • With time disruptorsmayalsobecomethemselvesdominant • Vertical Restraints • MFNs, Resalepricemaintenance, territorial restrictions, quotas on internet sales, etc.

  12. Challenges to Competition Agencies • Internationalco-operation • Disruptiveinnovationbringsnew and complexissues • As the geographic scope of online platforms expands, and the use of Big Data and algorithms increases, the nature of dominance and the scope of remedies to abusive behaviour will extend beyond the jurisdiction of a single competition authority • Benefits in sharingexperiencesandknowledge • Not only challenges for competition but also taxation, consumer protection, privacy issues, legal dispute settlement and product regulation (need for a global concerted effort and new thinking)

  13. Big data • Exponential growth of the digital economy  rise of business models based on the collection and processing of “Big Data”, with implications for competition authorities • Assess whether competition law is the appropriate tool for dealing with such issues • How to considerBig Data acrosstheenforcementspectrum? • BigData as anasset and as possiblebarrier to entry • Privacy as anelementofqualityofservice • As a basis for pricediscrimination • OECD: Hearing discussion on Big Data (November 2016)

  14. Big data The 4 V's of big data: • Volume: global data centre IP traffic is estimated to reach a forecasted value of 10.4 zettabytes in 2019. • Illustration: in order to store 10.4 zetabytes of data, every individual in the world would have to own 11 iPhones of 128 gigabytes. • Velocity: the speed at which firms access, process and analyse data is now approaching real time, allowing companies to forecast things as they happen - now-cast. • Example: Google's use of search queries to detect flu epidemics as they happen, as compared to traditional measures (such as patient visits to hospitals) that have time lags of 1-2 weeks.

  15. Big data Variety: data is collected from multiples sources, allowing firms to know costumers' age, gender, location, household composition, dietary habits, biometrics, preferences… Value: Big data allows companies to come up with product innovations, improve the efficiency of productive processes, forecast market trends, improve decision making and enhance consumer segmentation.

  16. Big data • Big data may have pro-competitive effects: New products and services, low prices, quality improvements, real-time supply, personalised recommendations, customised products, consumer information • But may also be a new source of market power due to • Economies of scale and scope • Network effects • Data feedback loops (data – quality of service – more users – more data) • This raises concerns for both competition law enforcer and regulators: careful weighting of the efficiency effects against the risks of enhanced market power and privacy and consumer protection issues

  17. Big Data: Implications for competition law enforcement • Big Data does not always lead to market power or winner-takes-all outcomes: • Data is cheap to collect and faces decreasing returns to the number of observations • Large data-driven companies frequently compete vigorously across multiple products • But there may be risks to competition: • Data can be seen as an input or asset that can be used for anti-competitive strategies (e.g. unilateral conduct, foreclosure) • Data may be part of non-price dimensions of competition, such as quality and innovation

  18. Big data: Anti-competitive effects Anti-competitive mergers Unilateral conducts Collusion Exploitative Exclusionary • Share data to facilitate tacit collusion • Use of data to identify and displace potential competitors through pre-emptive mergers • Prevent access to essential data in order to foreclose the market • Use of algorithms to implement cartels • Data motivated acquisitions that reinforce market power • Use of data-mining to discriminate consumers • Data analysis to identify threats and block entry

  19. Data regulation and competition policy In the design of a regulatory framework, competition law enforcers share common goals with agencies for consumer protection and data protection agencies Big data fades the line separating demand from supply agents, as online users supply data that is consumed and monetised by companies – consumers as Prosumers The creation of standards may empower consumers with control over their own data and allow them to better understand the terms of the service However, standards must be carefully designed in order not to stifle innovation Rules on data portability may have positive or negative impacts on competition and data protection, depending on how they are designed

  20. Algorithms and collusion “Thetroubleis, it’snoteasy to knowexactlyhowthosealgorithmswork. Howthey’vedecidedwhat to show us, and what to hide. And yetthedecisionstheymakeaffectusall” MargaretheVerstager (2017), EU Commissioner for Competition

  21. Algorithms and collusion • A growing number of firms are using computer algorithms • to improve their pricing models, • to customise services, and • to predict market trends • The competitive landscape may change with the combination of data with technologically advanced tools such as pricing algorithms, programming tools and machine learning technology • OECD: Roundtable on Algorithms and Collusion (June 2017) to discuss some of the new challenges raised by algorithms

  22. Artificial Intelligence Algorithms and collusion Machine Learning Deep Learning • Artificial intelligence • Detailed algorithms that mimic human intelligence • Machine learning • Algorithms that iteratively learn from data • Deep learning • Artificial neural networks that replicate the activity of human neurons…

  23. Algorithms and collusion Applications of Algorithms Business Consumers Government Consumer information Predictive analytics Crime detection Decision-making optimisation Process optimisation Determine fines and sentences Increasebuyer power Increasesupplier power Positive impact on static and dynamic efficiency

  24. Algorithms and collusion Pro-competitive business applications of algorithms Predictive analytics Optimisation of business processes Product innovation Supply-chain optimisation Dynamic pricing Risk management Price differentiation Target ads Product Customisation Recommendations Fraud prevention

  25. Algorithms and collusion Algorithmic collusion consists in any form of anti-competitive agreement or coordination among competing firms that is facilitated or implemented through means of automated systems.

  26. Algorithms and collusion • For collusion to besustainable: • Common understanding/Common policy • Internal Stability (monitoring deviations, punishment) • External Stability • Some marketconditionsfavourcollusion • Algorithms may increase the likelihood of collusion without the need for explicit communication, even in scenarios where only explicit collusion used to be possible • And they do itwithouthumanbiases…

  27. Algorithms and collusion Legend: + positive impact; - negative impact; 0 neutral impact; ± ambiguous impact

  28. Algorithms and collusion • In a perfectly transparent market where firms interact repeatedly, when the retaliation lag tends to zero, collusion can always be sustained as an equilibrium strategy. • Intuition: • If markets are transparent and companies react instantaneously to any deviation, the payoff from deviation is zero. • The speed and enhanced transparency of algorithms may allow for collusive outcomes, which may tacitly arise, in oligopolies with a greater number of players than we are normally used to.

  29. “Meeting of algorithms” – examples

  30. Summary of the risks of algorithmic collusion Algorithms Change market characteristics Govern collusive structures Common policy: Signal & negotiate Transparency Monitor & punish High frequency trading Optimise joint profits Coordinate common policy Increase likelihood of collusion Replace explicit communication Tacit collusion

  31. Competition enforcement challenges • Horizontal agreement • If competition agencies find evidence that algorithms were used to somehow negotiate prices, or allocate markets, the current antitrust rules can be enforced as in a brick and mortar world • Tacit collusion • Algorithms may however facilitate tacit collusion, where agents are operating autonomously, recognizing their mutual interdependence in the market – conscious parallelism • Can they do anything about it?

  32. Competition enforcement challenges • Facilitatingpractices • Although more difficult to prove, aninfringement can alsobeestablished in mostjusrisdictionsifthe use ofalgorithmsisfound to havebeenused to facilitatecollusion, i.e. theconsciousparallelism, amounting to anillicitconcertedpractice. • “Plus factors” (communication, exchange of sensitive information, signalling, manipulationthroughthesharingofdata…) • Challenges: understand how the technology works and how the algorithm can facilitate or support the main antitrust infringement • PureTacitCollusion • Normallywouldnotresult in any antitrust infringement • Should jurisdictions review their approach towards tacit collusion? • Enforcement gap?

  33. Competition enforcement challenges • Concept of agreement • Can a “meeting of algorithms” amount to an anti-competitive agreement? • Liability • Who is liable for the decisions and actions of algorithms? • Creators (programmers, third party centres) • Users (managers, commercials) • Benefiters (shareholders, managers, competitors) • Weak link between the agent (algorithm) and the principal (human being) • Defining a benchmark for illegality requires assessing whether any illegal action could have been anticipated or predetermined

  34. Alternative approaches

  35. Regulating algorithms? • Social media that results in “echo chambers” • Product recommendations based on past purchases • Content-control software to block specific information • Manipulation of feedback scores • Manipulation of rankings in search engines’ results • Automatic collection of personal data for target ads • Collection, use and share of information protected by IP rights, such as music and video • Price and product discrimination based on social characteristics

  36. Regulating algorithms? Can the risks of algorithmic selection be eroded through “algorithmic competition”? Marketfailures?

  37. Regulating algorithms? • Increasingalgorithmictransparency and accountability? • New FTC Office of Technology Research and Investigation responsible for studying algorithmic transparency • Algorithmic compliance with data protection and antitrust laws by design?

  38. Regulating algorithms? • Price regulation • Restrain the ability of firms to set high prices, regardless of whether they are achieved through explicit collusion, tacit coordination or other means Problems to competition from regulating • Reduces incentives to innovate or to supply high-quality products • Creates a focal point for collusion • Creates barriers to new market entrants

  39. Regulating algorithms? • Market design regulation • Change the structural characteristics of digital markets to make tacit collusion unstable • To decrease market transparency and frequency of interaction • E.g.: • impose restrictions on the information that can be published online or lags on price adjustments • require companies to compromise to any new offer for a minimum time period Problems to competition from regulating • Limits the information available to consumers • Restricts the ability of firms to adjust strategies fast and efficiently

  40. Regulations to prevent algorithmic collusion? • Rules on algorithm design • Force programmers to comply with competition principles in the design of algorithms – compliance by design • E.g.: algorithms could be programed not to react to most recent changes in prices or to ignore price variations of individual companies, while still accounting for average prices in the industry • Problems to competition from regulating • Limits the ability of firms to adjust strategies efficiently • Raises entry costs by forcing firms to comply by design

  41. More information & documents www.oecd.org/daf/competition

  42. Thank you!antonio.gomes@oecd.org

More Related